CONTAP-500025: ONTAP may panic with "Resource large_mbuf_flowcontrol had 100% stranded capacity" when ONTAP S3 clients are slow or unresponsive during connection shutdown

Issue

In rare conditions, an ONTAP node serving S3 requests can panic when many client TCP connections stop making progress during connection shutdown. The affected connections can remain in the FIN_WAIT_1 state while retaining socket send-buffer resources, eventually causing resource exhaustion.
When this occurs, the node may panic with a message similar to:
PANIC: Resource large_mbuf_flowcontrol had 100% stranded capacity for 720 seconds
This has been observed when S3 clients become slow or unresponsive: they stop reading data, advertise zero receive windows, and fail to acknowledge connection shutdown after ONTAP initiates a normal TCP close. With a large number of such stalled connections, each holding buffered data in its send queue, network memory buffers (mbufs) accumulate until the large_mbuf_flowcontrol resource pool is exhausted and nodewatchdog triggers a panic.